Performance of rotavirus vaccines in developed and developing countries

Introduction

Rotavirus diarrhea causes over 500,000 deaths and 2.4 million hospitalizations globally each year among children less than 5 years of age.^1–3 Because of the tremendous global burden of rotavirus, the World Health Organization (WHO) has prioritized vaccine development and introduction to control this disease. In recent years, clinical trials of two live oral rotavirus vaccines, RotaTeq® (RV5; Merck Vaccines, Whitehouse Station, NJ) and Rotarix® (RV1; GlaxoSmithKline Biologicals, Rixensart, Belgium) have demonstrated good efficacy against severe rotavirus disease and a reassuring safety profile in developed and middle income countries.^4–7 On the basis of existing data on the performance of these vaccines, WHO recommends the global introduction of these vaccines in national immunization programs worldwide.⁸ In 2006, WHO strongly recommended the inclusion of these new rotavirus vaccines into national immunization programs of countries in the Americas and Europe on the basis of pivotal clinical trials from these regions.⁹ In April 2009, WHO extended this recommendation to all regions of the world after review of clinical trial data from Africa and Asia, and postlicensure data from the Americas.⁸

The success of these vaccines will in part depend on their efficacy in the poorest settings with the highest burden of severe rotavirus disease. In this review, we summarize the experience of the current and previous candidate rotavirus vaccines, with emphasis on an examination of their performance in developed versus developing countries. Our primary interest was to understand if variations in vaccine efficacy exist between the two populations that are independent of vaccine type, study design, and outcome measures of efficacy. For rotavirus vaccines to have an optimal public health impact, sustained protection should be attained through first 2 to 3 years of life, when a majority of the severe rotavirus disease occurs worldwide.¹⁰ Thus, as a secondary objective we assessed for potential waning of immunity by restricting analysis to efficacy and effectiveness studies assessing duration of protection after rotavirus vaccination.

General Background on Rotaviruses: Factors Relevant to Vaccine Development and Introduction

Human rotavirus was first isolated by Ruth Bishop in 1973 from epithelial cells of the small intestine of children with diarrhea.¹¹ Rotavirus has an 11-segmented double-stranded RNA genome that is surrounded by three protein shells: a core, an inner capsid, and outer capsid.¹² Each RNA gene segment encodes a structural (VP1-VP4, VP6, VP7) or nonstructural protein(s) (NSP1-NSP4), except for gene 11 which codes for both NSP5 and NSP6. Two structural proteins, VP7 (a glycoprotein—G protein) and VP4 (a protease-cleaved protein—P protein), forming the outer shell of the rotaviruses are used for rotavirus strain characterization. The two gene segments that encode VP4 and VP7 can segregate independently in vivo and in vitro, and theoretically lead to more than 100 different G and P protein combinations, however four combinations (P[8]G1, P[4]G2, P[8]G3, P[8]G4) are responsible for >80% of the global burden of severe rotavirus disease.¹³ However, strains can vary from year to year, and by region; in addition, periodic emergence of novel strains (G9 and G12) have also been reported.

The mechanisms and duration of protection against rotavirus infection are incompletely understood, however, it is suspected that clinical protection likely involves mucosal (intestinal) and systemic antibody response as well as cell-mediated immunity.¹⁴ Evidence suggests that intestinal (jejunal) IgA is possibly one of the more important mechanisms for long-term protection against rotavirus infection.¹⁴ With regard to potential protection from systemic antibodies, immune response has been hypothesized to partly depend on neutralizing epitopes VP7 and VP4 on the viral capsid.^14,15 As such, the antigen composition of the current and candidate rotavirus vaccines are typically described on the basis of these VP7 (G-type) and VP4 (P-type) proteins on the vaccine strain. Does natural rotavirus infection with one strain confer cross-protection (i.e., heterotypic) against other antigens? If neutralizing epitopes VP7 and VP4 were the predominant modulators of immunity, for heterotypic protection to occur, strains that cause initial infections would need to share epitopes with strains that they will eventually protect against. The presence of a few predominant circulating strains globally, often over many years, argues against the serotype-specific neutralizing antibodies as the exclusive mechanism of protection—that is, under exclusive serotype-specific protection, selective pressure would have lead to an apparent antigenic drift and frequent changes in serotype prevalence. In addition, many observational studies and vaccine trials support the presence of heterotypic protection after primary infection.^10,14 However, it has been observed that the first infection with rotavirus elicits a predominantly homotypic, serum-neutralizing antibody response to the virus, and subsequent infections (asymptomatic and symptomatic) elicit a broader, heterotypic response.^16,17

The current and previous candidate rotavirus vaccines can be categorized in two groups on the basis of their development (Table 1):^18,19

1. first generation vaccines: these early rotavirus vaccines were single-strain animal strains that were naturally attenuated in that they did not cause clinical disease in humans but conferred protection against subsequent infection with human rotavirus strains;

2. second generation vaccines: these rotavirus vaccines developed subsequent to the first generation vaccines include two subgroups: (a) human-animal reassortants: these reassortants contain the backbone of an animal strain which incorporate additional genes from human strains by capitalizing on the viruses' ability to reassort in vitro. The development process involves combining naturally attenuated animal strains in cell culture systems with human rotavirus strains. Subsequently, reaassortant vaccine strains are recovered that contain either the VP7 or VP4 gene from a human strain, and the remaining 10 rotavirus genes from the naturally attenuated animal virus strain. (b) human strains: these are vaccines developed through attenuation of human rotavirus strains.

Results

Discussion

Our review highlights that both first and second generation rotavirus vaccines have largely demonstrated high efficacy (∼73% and ∼85%, respectively) against severe rotavirus diarrhea in industrialized nations. In poor settings with the greatest rotavirus disease burden, however, the immune response and clinical protection from rotavirus vaccines has been fraught with challenges.^18,19,69 Efficacy of the first generation rotavirus vaccines was poor (∼20%), however, substantial experience was gained from these early vaccines which led to strategic improvements in the development and application of second generation vaccines. These improvements and an aggressive research agenda led to success of rotavirus vaccines in developing country settings, where the second generation vaccines have shown to provide efficacy of ∼51% to 66% protection against severe rotavirus in clinical trials and postlicensure studies. Thus, while rotavirus vaccines still continue to be less effective in the most impoverished settings compared with rich populations, efficacy of the current vaccines has substantially improved in the past decade, possibly due to improvements in vaccine type, study design, age at vaccination, number of doses, or outcome measures studied. Moreover, despite the lower efficacy, the trials have unequivocally demonstrated that the absolute reduction in severe disease is greater in developing country settings than more affluent settings with higher efficacy.⁴⁸

Immune responses to rotavirus vaccination also vary between infants in developed and industrialized settings, with a trend of reduced immunogenicity in countries with the lowest income level.^70,71 While immunogenicity data do not directly correlate with efficacy, they may offer some further evidence of impaired immune response in the children from the poorest regions. For example, after immunization with RV1, infants in high income countries, where the pooled efficacy of second generation vaccines was 85%, had substantially higher concentrations of IgA antibodies (mean titers = 206 U/mL; seroconversion = 86%) compared with vaccinated infants in low-income countries (mean titers = 68 U/mL; serocoversion = 63%), where pooled efficacy was 51%.⁷¹ These variations in immune response are not surprising given the differences in epidemiology of rotavirus disease between the developing and developed countries, and could possibly reflect differences in force of infection, host characteristics (e.g., level of nourishment, breastfeeding), circulating strain patterns, and presence of other enteric pathogens.¹⁸ Rotavirus disease in poor tropical countries is often year round, likely reflecting the higher force of infection. Although immunogenecity may not perfectly correlate with efficacy, experience from clinical trials with first and second generation candidate rotavirus vaccines suggests that field performance of these vaccines is also less optimal in developing countries than in more affluent settings. Several findings from these efficacy studies in developing countries could shed light on future use of currently licensed vaccines in these settings. Our review highlights the difficulties in comparability of study results, particularly for the first generation vaccines, due to differences in study design, such as number of doses administered, age at vaccine administration, outcome measures of severity, and time of followup. While these differences in study design could have partly explained differences in observed efficacy between developed and developing regions, the consistency of the decrease in efficacy across multiple studies and oral vaccines supports the contention that host factors among children in developing countries are likely to play a role in lower efficacy.

Several recent reviews have explored potential explanations for the impaired immune response in poor settings in hopes of identifying strategies to improve protection against severe rotavirus disease among children who would most benefit the vaccine.^19,69–72 Factors that distinguish children in developing countries from industrialized nations have posed challenges to take of rotavirus vaccines in developing countries and oral vaccines for other diseases such as cholera,^73–75 typhoid⁷⁶ and polio,^77,78 and thus closer attention is warranted to identify modifiable factors that could enhance the performance of these vaccines in challenging settings.⁷¹ If these barriers are inherent and cannot be bypassed, alternative strategies may warrant consideration including schedule modifications, increasing vaccine titers, increase in number of doses, and inclusion of broader antigens so that the full lifesaving potential of these interventions is realized. Ongoing clinical development of other live oral vaccines with different strains (i.e., neonatal strains, rhesus and bovine strain variants) as well as parenteral vaccines could also potentially fulfill the promise of vaccines that are more affordable. Whether these candidate vaccines perform better in poor settings remains to be seen.^70,79,80

In both developing and developed countries, some proportion (30–40%) of severe rotavirus disease occurs between 12–36 months of age.¹⁰ Thus, for vaccines to have a substantial public health impact, they should ideally provide protection against severe rotavirus disease through 2–3 years of life. Our review indicates that in industrialized countries the second generation vaccines provide sustained protection against severe rotavirus disease through these important early years. However, clinical trial data on duration of protection in developing countries are yet to be published. The postlicensure evidence indicates a decrease in efficacy may very well be a possibility. The suggestion has been made that acquisition of immunity from natural infection may potentially explain decreasing efficacy with age. Two explanations argue against this theory. First, in a randomly vaccinated population, no biological reason exists to suspect a different rate of exposure (and acquisition of immunity) to natural infection between susceptible vaccinated and unvaccinated groups. Thus, incidence of rotavirus disease by increasing age should be similar between susceptible unvaccinated children and those who are vaccinated but do not mount immunity (i.e., susceptible vaccine failures). In absence of waning immunity, the reduction in disease incidence attributable to vaccination should be stable with age. Vaccine efficacy is computed on the basis of difference in disease incidence between unvaccinated and vaccinated, thus making it unlikely that efficacy estimates would decrease with age. Second, we would have observed reductions in efficacy during the second and third season after vaccination in trials from industrialized countries if, hypothetically, vaccine failures were less likely to mount immunity from natural infection compared with unvaccinated persons. Because immunity was sustained in these trials from industrialized countries for 3 years but diminished some in developing countries after 1 year suggests that the observed reduction in efficacy with age is likely to be related to waning of immunity.

In cohort studies following children with natural infection, although recurrent episodes of severe disease are unlikely, repeat infections do occur and suggests that natural protection may be short-lived or incomplete.^81–84 This important observation and the finding that repeated infections are more likely to induce a heterotypic response have provided support for use of two or more doses of an attenuated vaccine.^16,17 Clinical protection after rotavirus infection likely involves mucosal (intestinal) and systemic antibody responses as well as the cell-mediated immune system. Duration of intestinal immunity is shorter than systemic because intestinal antibodies decay at a rate faster than serum antibody titer decreases.¹⁴ Moreover, in developing countries, IgA titers are significantly lower than titers in industrialized countries.⁷¹ Although the immune correlates of protection from rotavirus infection and disease are not fully understood, taken together with the evidence that intestinal immunity is an important mechanism of protection,¹⁴ the lower systemic immune response in poor regions might provide one possible explanation for the shorter duration of protection in developing countries. The incomplete understanding of the mechanisms and duration of protection against rotavirus infection warrant further research so that effectiveness of efforts to reduce rotavirus illness can be maximized and development of other potential vaccine strategies can be realized.

We tried to minimize sources of heterogeneity related to differences in vaccine type and outcome measure in the different vaccine trials. This allowed us to better determine if inherent differences between children as well as the environmental conditions in industrialized and developing countries contribute to variation in vaccine performance between the two populations. It is certain that a gradient of poverty exists between countries we classified as “developing.” This gradient would also likely reflect in the efficacy estimates. Although a finer grouping of country classification would be optimum, the limited efficacy data by region did not allow us to further differentiate the developing countries. Such a distinction would be particularly useful for the large RV1 clinical trial from 10 Latin American countries.⁴ Of the 15,183 infants from 10 Latin American countries, 60% were from upper-middle income countries (Mexico, Panama, Venezuela, Brazil, Argentina and Chile) and 29% were low-middle income (Honduras, Colombia, and Dominican Republic) and 11% from a low income country (Nicaragua). The authors of this study presented pooled estimates from all 10 countries, and a breakdown of these efficacy estimates by country has not been published. Because a majority of the participants were from upper-middle income countries, we included this study in the industrialized region, as we surmised that the estimates are more likely to reflect the higher socioeconomic group rather than lowest socioeconomic group represented by the single country of Nicaragua. We suspect that a gradient of efficacy likely exists between countries in this region which would be useful to know for fully understanding the reasons for the location-dependent variation in protection.

In summary, substantial advancements towards the prevention of rotavirus disease have occurred in the three decades since the initial discovery of rotavirus. Through persistent research and advocacy, rotavirus is now accepted as a leading preventable cause of childhood morbidity and mortality worldwide. The path to vaccine development and implementation has been tortuous, with a major setback in 1999 due to an unexpected association between the highly effective RRV-TV vaccine and intussusception. However, an energetic research agenda and global public-private collaboration led to further development and testing of novel and safe strategies for rotavirus disease prevention. The current rotavirus vaccines have proven to be effective in a variety of developed and developing country settings, and have received a global recommendation for introduction by WHO. Even though vaccines have demonstrated somewhat lower efficacy and effectiveness in low income settings, they are still likely to have a substantial public health impact because of the higher background rates of severe rotavirus disease in these populations. Room for improvement in efficacy exists in the less affluent settings, and thus, to optimize the full life-saving potential of these vaccines, future efforts should focus on ways to improve immune response in challenging populations, and ensuring that demand and availability of vaccines is sufficient to reach those at the greatest risk of rotavirus disease.

Materials and Methods

Our review was restricted to clinical trials of vaccine efficacy against rotavirus disease and studies of postlicensure vaccine effectiveness.

To minimize differences in vaccine type, we categorized the current and previous candidate vaccines in two groups, as described above (Table 1). Uniformity in outcome measure is also an important consideration when assessing potential differences in vaccine performance by setting. Rotavirus vaccines are more efficient in preventing severe rotavirus disease than milder disease.^18,19 Thus, to minimize variation in protection observed in clinical trials as a result of differences in outcome measured (i.e., severity of rotavirus disease), we restricted our analysis to efficacy against ‘severe rotavirus disease.’ To the extent possible, we report efficacy estimates against severe rotavirus disease, as defined by severity score ≥11 on 20-point Vesikari scale or overnight admission to the hospital. The Vesikari scale is a 20-point scoring system that rates severity of rotavirus diarrhea on the basis of clinical symptoms and treatment.⁸⁵ When these outcome measures were not applied, we report efficacy estimates and the clinical parameter of severity presented by the authors.

Because our primary outcome of interest was to determine efficacy estimates of rotavirus vaccine in poor versus industrialized settings, we stratified data in two geographical regions on basis of World Bank data on the per capita income for the country where the study was conducted⁸⁶—countries with high and high-middle income economies were considered industrialized, and those with low- and low-middle income economies were considered developing. Studies from the US Native American Reservations were grouped in the low-income category because of the low-socioeconomic, rural nature of the population.¹⁹ For each group of vaccines, we calculated region specific pooled estimates and 95% confidence intervals (CI) for the vaccine efficacy parameters using random effects models, or fixed effects when appropriate (StatsDirect version 2.5.7, StatsDirect Ltd., Cheshire, England). For each of the randomized control trials, the relative risk of disease and 95% confidence intervals (CI) among vaccinated and unvaccinated study participants was calculated. To obtain region specific estimates of vaccine efficacy, estimates of risk ratios were pooled from all studies by vaccine type (first versus second generation) and region (developing versus industrialized). Heterogeneity was reduced by restricting pooling to vaccine type and region, and using severe disease as outcome of interest. To address additional variation across studies, the random effects model was applied using the DerSimonian and Laird method to estimate the pooled risk ratio for each strata of interest. For the postlicensure case-control studies of the second-generation vaccines, the odds ratio of vaccination between cases and controls and 95% CI for each study by region were used to estimate a pooled estimate using a fixed effects model. Stratum weights were calculated as the inverse of the variance and subsequently summed to calculate a pooled estimate using a weighted mean.